Validation of the prognostic signature

The validation of the gene signature was accomplished on independent data sets. Gene expression data from different data sets were adjusted individually by subtracting the median expression value across the samples. To further refine this model and to sub-stratify the predicted outcomes, Compound Covariate Predictor (CCP) was utilized as a class prediction algorithm [50]. The robustness was estimated by the misclassification rate that was determined during the leave-one-out cross-validation (LOOCV) in the training data set.

Kaplan-Meier survival analyses were performed after patient classification into two risk groups, and Chi-square (χ²) and log-rank tests were used to evaluate the survival risk between two predicted subgroups of patients. The univariate and multivariate Cox proportional hazard regression analyses were used to evaluate independent prognostic factors associated with survival. Gene signature, stage, smoking, gender, and age were employed as covariates.